br Fig Intracellular calreticulin and heat shock proteins

Fig. 3. Intracellular calreticulin and heat shock proteins as measured by immunofluorescent microscopy are similarly increased after treat-ment with cisplatin or oxaliplatin in HNSCC cell lines. Two cell lines were grown on chamber slides and treated with cisplatin or oxaliplatin for 48 h, stained with Hoechst to label nuclei, then fixed and stained with fluorescent anti-bodies to calreticulin or HSP70. Slides were then analyzed by confocal microscopy with identical laser settings across samples. Micrographs shown are representative of triplicate samples from two independent experiments.
anti-PD-1 (Fig. 6C). However, these differences did not reach statistical significance, though they were nearly significant for calreticulin (p = 0.06, one-way ANOVA).
Platinum chemotherapy drugs may differentially affect CD8+ T cells, but not myeloid cells, when combined with PD-1 blockade
Tumors from treated animals were also analyzed for immune cell subsets by flow cytometry. The number of CD8+ T ML385 was sig-nificantly higher in tumors from animals treated with anti-PD-1 + ox-aliplatin versus cisplatin (Fig. 6D). However, the percentage of CD8+ T cells with CD137 (4-1BB costimulatory molecule) expression was no-tably higher in tumors from animals treated with anti-PD-1 antibody; an effect that was lost with the addition of oxaliplatin but not cisplatin. Expression of CD107a, a marker of activation on CD8+ T cells, was increased with anti-PD-1 but was reduced back to baseline levels when either chemotherapy drug was added. Levels of dendritic cells and myeloid-derived suppressor cells were not significantly different among the treatment groups (data not shown). These results suggest that cis-platin and oxaliplatin, when paired with PD-1 blockade, have no ap-preciable effects on myeloid cells but may affect CD8+ T cells. Ox-aliplatin may be more likely to increase the overall number of CD8+ T cells in tumors, whereas cisplatin may be less likely to impair some aspects of T cell function, such as CD137 costimulation.
Discussion
Previous preclinical studies involving non-HNSCC cancer types suggest that oxaliplatin is a bona fide ICD inducer, but other platinum drugs are not [6,9,12,13]. In particular, cisplatin is thought to be a suboptimal ICD inducer due to its failure to induce calreticulin exposure on the cell surface [6,9,12,13]. It should be noted that the doses of cisplatin used in those studies were supraphysiologic, and time points selected were short. In our studies, we carefully titrated the doses to determine a moderate dose capable of killing 40% of cells over a 72 h period, and all of these doses were within the range of serum con-centrations seen in patients treated with cisplatin for HNSCC [24]. At 
these doses, we noted increased calreticulin both in the cells and at the cell surface in multiple cell lines, with no appreciable differences be-tween cisplatin and oxaliplatin. Our experiments failed to show all the classic markers for ICD, including release of HMGB1 and ATP into the supernatants, likely because we stopped the experiments at 48 h when cell death was negligible. However, even high doses (LD90) failed to induce robust ICD in our mouse vaccination experiments. We posit that the doses of cisplatin or oxaliplatin monotherapy that would be re-quired to induce classic ICD in experimental settings would be much higher than achievable serum concentrations in patients, and thus prior studies suggesting superior ICD with oxaliplatin should not be used to justify choosing this drug over cisplatin in trials of chemoimmu-notherapy.
Mouse vaccination experiments, which are the gold standard for demonstration of ICD [21], suggested that cisplatin and oxaliplatin may both be relatively weak ICD inducers. In a syngeneic mouse model, cisplatin and oxaliplatin produced a virtually identical tumor growth delay, whether used alone or in combination with PD-1 blockade. Our in vivo mouse experiments also suggest that cisplatin and oxaliplatin may affect the number and function of CD8+ T cells in different ways: oxaliplatin significantly increased the number of CD8+ cells, whereas cisplatin was less likely to negate the anti-PD-1-induced increase in the proportion of CD137-positive cells. These differential effects of cisplatin and oxaliplatin are somewhat preliminary based on the small number of animals, but the results were statistically significant, and suggest in-triguing avenues for further development.